Optically tunable microcavity in a planar photonic crystal silicon waveguide buried in oxide
Iwan Märki, Martin Salt, and Hans Peter Herzig
Institute of Microtechnology, University of Neuchâtel, Rue A.-L. Breguet 2, CH-2000 Neuchâtel, Switzerland
Ross Stanley
Centre Suisse d’Electronique et de Microtechnique SA, Rue Jaquet-Droz 1, CH-2007 Neuchâtel, Switzerland
L. El Melhaoui, P. Lyan, and J. M. Fedeli
Laboratoire d’Electronique de Technologie de l’Information, CEA-Grenoble, 17 Rue des Martyrs, 38 054 Grenoble Cedex 9, France
We present all-optical tuning and switching of a microcavity inside a two-dimensional photonic crystal wave- guide. The photonic crystal structure is fabricated in silicon-on-insulator using complementary metal-oxide semiconductor processing techniques based on deep ultraviolet lithography and is completely buried in a silicon dioxide cladding that provides protection from the environment. By focusing a laser onto the micro- cavity region, both a thermal and a plasma dispersion effect are generated, allowing tuning and fast modu- lation of the in-plane transmission. By means of the temporal characteristics of the in-plane transmission, we experimentally identify a slower thermal and a fast plasma dispersion effect with modulation band- widths of the order of several100 kHz and up to the gigahertz level, respectively.
The ability to control light propagation on a wave- length scale by use of silicon planar photonic crystal structures is well known. Planar photonic crystal waveguides, bends, and microcavities are considered basic building blocks for applications such as micro- lasers, filters, multiplexers, and optical switches.1–4 It is possible to tune or switch photonic crystal de- vices in various ways, for example, by control of temperature,5 refractive index change using liquid crystals,6 control of free charge carrier density,7 or nonlinear material effects.8 Active photonic crystal devices allow further miniaturization of integrated optical circuits with active functionality. High-Qmi- crocavities in planar photonic crystals exhibit highly localized fields and narrow transmission bands.9 Even a small perturbation of the localized field can change the transmission properties of the microcav- ity. The relatively weak nonlinear properties of sili- con can be used in these structures for low-power modulation and switching.10–12
Until now most of the work done on tuning and modulating silicon photonic crystal devices has been based on an electrically controlled thermo-optical ef- fect. In this work, we present an optically tunable mi- crocavity inside a photonic crystal waveguide. We show experimentally that tuning the resonant wave- length and modulating the transmission signal is possible by using a laser focused onto the microcavity region, generating both thermal and plasma disper- sion effects. As silicon membranes are delicate and thus sensitive to the environment, our waveguide structures are fully buried in a silicon dioxide clad- ding being protected and more practical for many ap- plications, notwithstanding the smaller refractive in- dex contrast.
The design of the in-plane resonant cavity is shown in Fig. 1. The photonic crystal consists of a triangular array of cylindrical holes in a thin silicon layer. The silicon slab is completely buried in a silicon dioxide cladding and is therefore protected from the environ- ment. The in-plane resonant cavity is formed by two identical one-dimensional Bragg reflectors inside the photonic crystal waveguide and has a length of
⬃330 nm. The cavity structure is designed to yield a highly localized optical field inside the cavity and high in-plane transmission for a narrow frequency band. For ease of fabrication, we have kept the de- sign as simple as possible while accepting lower per- formance for the cavity. Thus, the holes of the one- dimensional Bragg reflectors are all equally spaced
Fig. 1. In-plane photonic crystal microcavity structure buried in silicon dioxide (PhC perioda= 490 nm, hole diam- eter d= 350 nm, core thickness t= 238 nm, bottom oxide thickness s⬇2000 nm, upper cladding thickness c
⬇700 nm). The microcavity inside the photonic crystal waveguide is formed by two five-hole Bragg reflectors (hole dimensions b1 = 325 nm and b2 = 180 nm, period p
= 373 nm).
Published in Optics Letters 31, issue 4, 513-515, 2006
which should be used for any reference to this work 1
and have the same dimensions. The designed photo- nic crystal microcavity structure was fabricated by the Laboratoire d’Electronique de Technologie de l’Information, France, using deep ultraviolet lithog- raphy at a wavelength of 193 nm. After the lithogra- phy process, the patterns in the photoresist are transferred into the silicon layer of a silicon-insulator wafer by use of HBr dry etching technology via a sili- con dioxide hard mask. As a final step, the etched structures in the silicon are then covered with an ox- ide layer.
The transmission measurements are performed us- ing a tunable laser source. By means of an aspheric lens, TE-polarized light (in-plane polarization) is in- jected into a 10m wide silicon waveguide. The light is guided through a taper, which reduces the wave- guide width to 0.5m, to the photonic crystal wave- guide. For a narrow frequency band the light propa- gating in the photonic crystal waveguide couples into the resonant cavity, where it is highly localized and eventually transmitted. With a second, identical taper the transmitted light is then guided to the exit of the device, where it is collected by a microscope ob- jective and focused onto an InGaAs detector for mea- surements.
The measured transmission spectrum of the micro- cavity exhibits a resonance peak at a wavelength of 1.561m with an approximate maximum transmis- sion efficiency of 11%. The quality factor is of the or- der of 200. Further transmission and loss properties of the microcavity were discussed in Ref. 13.
At the transmission resonance the spectral fre- quency response of the microcavity is sensitive to the smallest perturbations of the region around the local- ized field. To induce a perturbation, a laser with a wavelength of = 532 nm and a maximum power of 30 mW is focused onto the center of the microcavity.
The light is strongly absorbed共⬃30%兲by the silicon
core of the waveguide, whereas the upper and lower oxide claddings are transparent for the control laser wavelength. Figure 2 shows the measured transmis- sion spectra through the microcavity [curve (1)] with- out and [curve (3)] with the control laser turned on.
The raw measurement data have been numerically smoothed to remove the periodic Fabry–Perot inter- ference (compare with the inset in Fig. 2), which is due to the reflection at the input and output wave- guide end facets. The resonant wavelength is red- shifted by 3 nm, and the transmission is attenuated by approximately 10% due to thermo-optical and plasma dispersion effects.
To distinguish the different physical effects we ob- serve the temporal response of the transmission sig- nal when a square modulation pulse of 12s is ap- plied to the control laser, shown in Fig. 3. A wavelength of = 1.564m, slightly above the reso- nant wavelength (the peak wavelength of the ther- mally shifted transmission spectrum, see Fig. 2), was chosen for the transmitted light to conveniently ob- serve both the thermo-optical and the plasma disper- sion effects in the modulated transmission signal.
The switching time of the control laser is of the order of 100 ns. Coinciding with the switch-on of the con- trol laser, we observe a small negative peak [position (2) in Fig. 3], which is due to the fast plasma disper- sion effect inducing a small negative index change and an increase in the absorption coefficient. Follow- ing, the transmission efficiency increases with a rise time of⬃5s as a result of the slower thermo-optical effect [position (3)], which induces a positive index change, shifting the resonant wavelength toward the wavelength of the transmission signal. At the switch- off of the probe laser a small positive peak is ob- served [position (4)]. This peak is due to the recombi- nation of the free carriers, which increases the refractive index and reduces the absorption coeffi- cient slightly. Sequentially, the transmission proper- ties return to the initial state with thermal relax- ation. For a better understanding of the present Fig. 2. Measured transmission spectra through the photo-
nic crystal microcavity (1) without and (3) with the control laser illumination共= 532 nm兲. Curves (2) and (4) approxi- mately indicate the change induced by the plasma disper- sion and thermo-optical effects, respectively. The arrow in- dicates the fixed wavelength for the temporal response measurements. The inset depicts the unfiltered transmis- sion spectrum with the periodic Fabry–Perot interference oscillations.
Fig. 3. (a) Measured temporal response of the transmis- sion signal with its switching characteristics [positions (1) to (4)]. The slight decrease of the signal in position (3) is most likely due to secondary thermal effects induced in the bulk silicon. (b) Square modulation pulse applied to the control laser.
2
effects the consecutive change in the transmission properties can be indicated in Fig. 2 according to the observed temporal response. The arrow in Fig. 2 marks the fixed wavelength of= 1.564m. The dot- ted curve (2) corresponding to position (2) in Fig. 3 approximately indicates the change induced by the plasma dispersion effect alone, whereas the dotted curve (4) corresponding to position (4) in Fig. 3 ap- proximately indicates the change due to the thermo- optical effect.
From these measurements, the size of the thermo- optical and plasma dispersion effects can be esti- mated. Three-dimensional simulations show that a change of ⌬n= + 0.05 in the refractive index of the silicon layer gives the required redshift. For silicon, dn/ dTis 1.86± 0.08⫻10−4/ K at 1.55m,14indicating a local temperature rise of 270° C due to the thermo- optical effect. The observed attenuation of approxi- mately 10% is due to the presence of free carriers generated by the plasma dispersion effect. Assuming an approximate effective cavity length of Q⫻Lcavity, the absorption coefficient is of the order of 15 cm−1. Based on the Drude model,15the induced charge den- sity is estimated to be ⌬N⬇1018cm−3, causing a re- fractive index change of approximately⌬n= −0.0007, which is more than one order of magnitude smaller than the thermo-optically induced index change. This explains the observed redshift in the transmission spectrum.
Furthermore, from Fig. 3, we estimate the thermo- optical maximum modulation frequency to be of the order of 200 kHz. Below this frequency a modulation depth of ⬃30% can be achieved (see Fig. 2). The plasma dispersion effect allows much greater modu- lation frequencies. We measured the transmitted sig- nal modulated at 50 MHz working above the thermal maximum modulation frequency. For modulation of the control laser an electro-optical resonant modula- tor was used, limiting the maximum modulation fre- quency to 50 MHz. In this case, the modulation depth is reduced to⬃10% owing to the weaker plasma dis- persion effect and to the diminished modulation depth of the control laser. A relaxation time for free carriers in a different light-confining structure on silicon has been measured to be of the order of 500 ps.16,17 Therefore, modulation frequencies in the gigahertz range should be possible.
A higher Q factor and transmission efficiency would allow a reduction in the power of the control laser and an increase in the maximum modulation frequency, improving the performance of the photonic crystal optical modulator. The differing temporal re- sponse of the thermal and plasma dispersion effects should allow simultaneous tuning and high-speed modulation in a limited range, increasing functional- ity. Based on the plasma dispersion effect, dense in- tegration of several structures not suffering from the thermal cross-talk encountered in present-day inte- grated optical devices becomes possible. With current technology one could envision the integration of the
control laser source onto the same chip, by stacking a wafer-based laser aligned with a wafer-based wave- guide. Alternatively, integration of the optical tuning and the modulation method into a completely planar solution by use of a different wavelength for the con- trol laser (slightly different from the wavelength of the transmission signal) could be realized with two intersecting resonant cavities. This configuration would permit modulation because of the free-carrier- induced change of the refractive index based on two- photon absorption.
In conclusion, we have demonstrated that high- speed optical switching and resonant wavelength tuning on the subwavelength scale are achievable with a relatively simple technique. The induced ther- mal and plasma dispersion effects have been clearly identified by means of the temporal characteristics of the transmission signal. These results indicate that an improved design based on the device presented could serve as a basic component of high-speed all- optical integrated circuits based on silicon photonic crystal structures with applications in, for example, high-bandwidth data routing.
I. Märki’s e-mail address is iwan.maerki@unine.ch.
References
1. A. Mekis, J. C. Chen, I. Kurland, S. Fan, P. Villeneuve, and J. D. Joannopoulos, Phys. Rev. Lett. 77, 3787 (1996).
2. S. Noda, M. Imada, M. Okano, S. Ogawa, M.
Mochizuki, and A. Chutinan, IEEE J. Quantum Electron.38, 726 (2002).
3. M. Lončar, T. Yoshie, A. Scherer, P. Gogna, and Y. Qiu, Appl. Phys. Lett.81, 2680 (2002).
4. M. Notomi, A. Shinya, S. Mitsugi, E. Kuramochi, and H.-Y. Ryu, Opt. Express12, 1551 (2004).
5. E. A. Camargo, H. M. H. Chong, and R. M. De La Rue, Opt. Express12, 588 (2004).
6. Ch. Schuller, F. Klopf, J. P. Reithmaier, M. Kamp, and A. Forchel, Appl. Phys. Lett.82, 2767 (2003).
7. S. W. Leonard, H. M. van Driel, J. Schilling, and R. B.
Wehrsporn, Phys. Rev. B66, 161102 (2002).
8. A. Locatelli, D. Modotto, D. Paloschi, and C. De Angelis, Opt. Commun.237, 97 (2004).
9. Y. Akahane, T. Asano, B.-S. Song, and S. Noda, Nature 425, 944 (2003).
10. V. R. Almeida, C. A. Barrios, R. R. Panepucci, and M.
Lipson, Nature31, 1081 (2004).
11. V. R. Almeida and M. Lipson, Opt. Lett. 29, 2387 (2004).
12. P. E. Barclay, K. Srinivasan, and O. Painter, Opt.
Express13, 801 (2005).
13. I. Märki, M. Salt, H. P. Herzig, R. Stanley, L. E.
Melhaoui, P. Lyan, and J. M. Fedeli, J. Appl. Phys.98, 013103 (2005).
14. G. Cocorullo and I. Rendina, Electron. Lett. 28, 83 (1992).
15. R. A. Soref and B. R. Bennet, IEEE J. Quantum Electron.QE-23, 123 (1987).
16. M. Lipson, Nanotechnology15, 622 (2004).
17. V. R. Almeida and M. Lipson, Opt. Lett. 29, 2387 (2004).
3
